lisa spiro - wordpress.com · offer customized instruction and feedback ... based on the act-r...
TRANSCRIPT
http://www.flickr.com/photos/hinkelstone/2435823037/
Lisa Spiro “IB in a Virtual World”
October 5, 2013
Timothy Dilich
“adaptive learning takes a sophisticated, data-driven, and in some cases, non-linear approach to instruction and remediation, adjusting to a learner’s interactions and demonstrated performance level and subsequently anticipating what types of content and resources learners’ need at a specific point in time to make progress.” (Learning to Adapt)
Clickers/ audience response systems (e.g. Contingent Pedagogies)
Games (e.g. Refraction) Textbooks Computer Adaptive Testing Intelligent tutoring systems/ cognitive tutors
“Adaptive learning may prove to be exactly what we need most right now – practices and tools that enable maximum learning gains for a diverse and broad array of students irrespective of their prior educational performance and preparation.”
(Josh Jarrett and Rahim Rajan, Gates Foundation) >> Gates recently issued an RFP for “Adaptive
Learning Market Acceleration Program”
“Computer adaptive learning systems are reductionist and primarily attend to those things that can be easily digitized and tested (math, science and reading). They fail to recognize that high quality learning environments are deeply relational, humanistic, creative, socially constructed, active and inquiry-oriented.” (Philip McRae)
I. Understanding adaptive learning systems
II. Case studies of 3 high profile adaptive learning systems
I. Carnegie Cognitive Tutor
II. Open Learning Initiative
III. Khan Academy
III. Making an informed
decision I. Benefits, challenges & risks
http://www.flickr.com/photos/electrichamster/96531540/
1. Conventional class with 30 students 2. Mastery learning: students master material
through formative tests, feedback and correction 3. Tutoring for 1-3 students, followed by formative
tests, feedback and correction
Type of Learning
Results
Conventional class
--
Mastery learning
1 std. deviation (average student above 84% of students in conventional class)
Tutoring 2 std. deviation (average student above 98% of students in conventional class)
How might we scale the effectiveness of tutoring without facing the high costs?
VanLehn (2011)
Type of Learning Results
Human Tutoring .79 standard deviation
Step-based intelligent tutoring systems
.76 standard deviation (sigma)
Two key differentiators of human tutors & intelligent tutoring systems (VanLehn): • Feedback: tutors help students identify errors and
correct them • Scaffolding: tutors help students think through
problems, take on increasing responsibility
interactive ≥ constructive > active > passive Interactive: dialoguing with peer or expert Constructive: producing outputs, e.g. concept map, hypothesis, questions Active: “doing something,” e.g. highlighting, pointing Passive: listening to a lecture
Via VanLehn (2011)
“computer software designed to simulate a human tutor’s behavior and guidance.” (ELI)
Based on research into artificial intelligence Provide exercises until mastery is demonstrated Offer customized instruction and feedback Collect detailed data about the learner Enable
Students to learn at their own pace
Instructors to monitor learning & intervene
Based on the ACT-R theory of knowledge (“learning by doing)
Grew out of Pittsburgh Urban Mathematics Project Algebra Tutor (early 1990s)
“Model tracing” tracks student’s steps in problem solving and provides feedback
“Knowledge tracing” monitors learning across problems to select next problems
Supports: Mastery learning
Differentiation of instruction
Integrated approach: text, software, training Recommendation: classes spend 2 days/week using the software 3 days/week in face-to-face class using active
learning techniques, e.g. problem solving, group work
Koedinger & Corbett, 2006
Out of 27 studies of Carnegie Learning Curricula and Cognitive Tutor, only 3 are fully eligible and 3 eligible with reservations
High standards for eligibility, e.g. randomized controlled trial
Found “mixed effects” on students’ math scores: one study showed statistically significant positive effect
on student performance
one (Geometry) a statistically significant negative effect
four showed indeterminate effects
U.S. Department of EducationWhat Works Clearinghouse. (2013). High School
Mathematics intervention report: Carnegie Learning Curricula and Cognitive Tutor®
Method: Randomized controlled effectiveness trial Involved 73 high schools and 74 middle schools Compared courses using Cognitive Tutor Algebra I (CTAI)
+ textbook to those using a traditional textbook
Findings: Use of CTAI led to “a significant positive effect in high schools in the
second year of implementation” (but not in the first year)
“equivalent to moving an algebra I student from the 50th to the 58th percentile”
Cf. Reich
Initiative launched by Carnegie Mellon U in 2002 Open educational project (free/ low-cost access) Focused on college and community college courses such
as statistics, chemistry, French & economics (18 to date) Incorporates embedded “mini-tutors”: simple versions
of cognitive tutors to provide feedback & assessment Collaboratively designed by subject experts, cognitive
scientists, human-computer interaction experts, &c Collects granular data on learning Provides feedback to learners and instructors
OLI
Lovett, Marsha, Oded Meyer, and Candace Thille. ”The Open Learning Initiative: Measuring the Effectiveness of the OLI Statistics Course in Accelerating Student Learning.” Journal of Interactive Media in Education 2008, no. 1 (2008).
OLI Accelerated
Average % Correct
Traditional Control
Average % Correct
Pre 55 Pre 50
Post 73 Post 53
Increase: 18 percentage points
Increase: 3 percentage points
Student performance on Comprehensive Assessment of Outcomes in a first [college-level] Statistics course
“students in OLI-Statistics took half the time to learn as much or more than their traditional counterparts”
“No significant difference” in learning outcomes Students had a slight preference for traditional Hybrid took ~ 25% less time to complete
Bowen, William G, et al. Interactive Learning Online at Public Universities: Evidence from Randomized Trials. Ithaka S+ R, 2012.
Provides free access to educational materials, including videos, exercises, and assessment tools
Founded 2006 High level of use
283 million views of YouTube channel
Incorporates adaptive techniques
Creating a profile of the learner
Adjusting difficulty of exercises based on prior responses
Offering detailed assessment data to learner & coach
Los Altos School District Blended learning model where Khan is just part of overall
approach that also includes manipulatives, projects, etc.
More value in exercises & data than videos (best as supplement)
Increased student motivation & ownership Envision Academy (Oakland) small pilot
Enabled teacher to do more 1:1 & small group interactions
Teacher could examine learner data (but needed ask why)
Students appreciated instant feedback, became more responsible for learning, and coached each other
SRI study due soon
“instructivist” approaches
Inaccuracies in videos (Khan Mystery Science Theater)
“do this then do this” pedagogy
Distracting people from deeper problems with education, such as inequity
Adapt Courseware Cerego Global CogBooks Jones & Bartlett
Learning LoudCloud McGraw-Hill Education
(LearnSmart) Open Learning
Initiative Smart Sparrow
Education Growth Advisors
Allow for personalized learning
More scalable than human tutors
Available at any time, any place
May be easier for reserved students to interact with
Create low-stakes opportunities to practice and build towards mastery
Provide detailed data on learning that the instructor can use to tailor teaching
Often require the learner to be motivated and self-directed
Work best in areas where there are clear answers (procedural/ fact-driven)
Expensive to develop
Time to integrate into education: technical infrastructure, curriculum, teacher training
Challenging for teachers to make sense of all the data
We lack sufficient understanding of effectiveness and need better approaches to evidence (US Dept. of Education, 2013)
Mechanizes learning
At odds with creativity, choice & social relationships
Diminishes teacher’s autonomy
Lack of clear research demonstrating benefits
Enforces rigidity and conformity Too much screen time?
Wikimedia Commons
Cost of adaptive learning systems are often higher than traditional textbooks
Do we want to outsource core
learning functions to private companies?
Invest in teachers rather than technology?
Lack of ownership of subscription-based resources Stacks of pound coins
Do we want to become the NSA of learning?
respect my privacy
“The purpose of school isn't to get people comfortable with life under constant observation. The endless efforts at data collection to capture what ‘works’ with learning have the potential to disrupt the learning they are trying to capture. Learning requires trust…” (Bill Fitzgerald)
Data that can be collected includes:
Social Security numbers
Disciplinary record
Whether a student is homeless
Concerns about who can access the data:
Employers?
Marketers?
inBloom responses:
Data is anonynmized
Data will not be sold
School districts determine what data will be collected
Most advocates of adaptive learning also emphasize the need for human teachers (cf. VanLehn, Meyer/OLI, Khan)
Adaptive learning: practice, hints, feedback, assessment, data on student performance
Human teacher: facilitating groups, helping students communicate, providing context, coaching, designing learning activities (but not grading tests)
Adaptive learning is typically used in a blended setting But I could see a scenario where fewer teachers are
required
Is this software consistent with the school mission and identity?
How will parents respond? Students? Teachers? What evidence is there that this technology works? How will the software be integrated into the curriculum? How will teachers will be trained to use the software? What are the current and ongoing costs? What happens
if the company goes out of business? Is there an adequate technical infrastructure (devices,
networking, etc) to support the software? Who has access to student data? How will that data be
protected?
Category Change
How class time is spent
Mix of computer-based interactions, individualized instruction & group work
Role of the teacher
Coach, learning designer, facilitator
Pace of learning
More according to the student’s needs, not the school calendar
you think adaptive learning systems could improve
learning you are concerned that adaptive learning systems
could undermine learning
you aren’t sure what to think
“If… each student has his or her own computer and he/she does an online course which adapts to the level and needs of the student, there will be a much greater degree of differentiation. They are all doing mathematics, but one student is doing statistics, another is starting to look at the arithmetic mean ... The teacher is still needed though, to clarify concepts, to set creative open-ended problems and to ‘steer the ship’.” (Dr Conrad Hughes, Director of Education in the International School of Geneva)
What impact might adaptive learning systems have on teaching and learning?
What potential benefits do you see to adaptive learning systems? What about disadvantages?
Are adaptive learning systems consistent with IB values? What strategy would you pursue with regards to
adaptive learning systems? How would you explain this strategy to your teachers? Parents? Students? Board members?
What are some effective approaches to differentiating instruction, with or without technology?
http://digitalscholarship.wordpress.com/
Bloom, Benjamin S. “The 2 Sigma Problem: The Search for Methods of Group Instruction as Effective as One-to-One Tutoring.” Educational Researcher 13, no. 6 (June 1, 1984): 4–16. doi:10.3102/0013189X013006004.
Education Growth Advisors, Learning to Adapt (2013) Fletcher, Seth. “How Big Data Is Taking Teachers Out of the
Lecturing Business: Scientific American.” http://www.scientificamerican.com/article.cfm?id=how-big-data-taking-teachers-out-lecturing-business.
Hanford, Emily, and Stephen Smith. “One Child at a Time: Custom Learning in the Digital Age.” American Radio Works, August 2013. http://americanradioworks.publicradio.org/features/personalized-learning/.
McRae, Phil. “Rebirth of the Teaching Machine through the Seduction of Data Analytics: This Time It’s Personal.” For the Love of Learning, April 29, 2013. http://www.joebower.org/2013/04/rebirth-of-teaching-machine-through.html.
US Department of Education, Expanding Evidence Approaches for Learning in a Digital World (2013)
VanLehn, Kurt. “The Relative Effectiveness of Human Tutoring, Intelligent Tutoring Systems, and Other Tutoring Systems.” Educational Psychologist 46, no. 4 (2011): 197–221.
What Works Clearinghouse